Rational Descriptive Classification of Duricrusts

EARTH SCIENCE JOURNAL Vol. 3, No. 2, 1969

RATIONAL DESCRIPTIVE CLASSIFICATION OF OURICRUSTS

G.H.DURY Department of Geography and Geology, The University of Wisconsin

Abstract The term duricrust appears to be extending itself to include calcareous, gypseous, and saline crusts, in addition to crusts composed dominantly of silica and/or of sesquioxides of iron and aluminium, with. -or without significant contents of dioxides of manganese or titanium. This latter group can be distinguished as duricrusts proper. Its nomenclature is highly confused, and its classification, in writings in the English language, defective. The relevant problems can be resolved, at least in considerable part, by the introduction, adaptation, and extension of modern terms current in tropical pedology, to give a descriptive classification free of genetic implications. When content of Si02, AbOg, and Fe20g is used as a primary basis for the classification of duricrusts proper, plots on a ternary diagram justify the recognition of seven named types in the fersiallitic range.

The nomenclature and c1assHicati.on .of duricrusts, as they have developed in English. are undeniably cDnfused: On a far smaller and much less complex scale, the situation resembles that faced by the authors of the 7th Approximation to the comprehensive classification .of soils (V.S. Department Df Agriculture, 1960). Perhaps a more apt comparison would be with the difficulties in the classification .of climate which Thomthwaite (1931. 1948) set out to resolve. Both in pedology and in climatology. the authors of the new classifications aimed at generally- applicable systems, relying on consistent criteria thrDughDut, and intrDducing new terminologies which relate to precise definitiDns and which are free from the unwanted imprecision and associations of previDus terminologies. Although it does not yet seem pOlssible to dO' as much for the naming and classification of duricrusts, the dubious and at times misleading connotations Df some of the terms in current use, plus seriDus deficiencies in what may be called the first apprDximative classificatiDn in English, point the need fDr an attempt to produce some kind .of taxonomic order. As is well known, the term duricrust was .originated by WODlnough ( 1927) for application tD surficial crusts in Australia. These had previously been variDusly described as . laterite, lateritic bauxite., bauxitic laterite, bauxite, concretionary ironstone, surface quartzite, and Desert Sandstone, according to assumed content .or origin and/or observed litholDgy and mineralDgy. The names listed cannot however be regarded as even a first appro~imation to a ratiDnal classification of duricrusts. Those in the laterite-bauxite range, quite apart frDm the uncertainties about chemical and mineralogic constitution which they entail, fail to make the necessary distinction between chemistry and mineralogy on the one hand and physical characteristics .on the other. The terms lateritic, bauxitic, &c., crusts would seem to be called fDr, even thDugh thick and well-cemented crusts can grade laterally into thin surface litters .or SUbSDil layers .of fragments, pisoliths, and nodules, and may wedge 0'ut altDgether. By nO' means all bauxites are sufficiently well cemented to' form crusts, and many appear to consist of transported material as opposed to residues in situ. Laterite, on anything like a strict definitiDn, need not be and usually is not indurated at all. The prevalent .obstacles to arriving at a strict definition of laterite need not concern us here (for discussion, see for

77 example Prescott and Pendleton, 1952; Maignien, 1966), since the term does not enter into the proposed classification. It is however worth observing that Buchanan's original laterite of 1807 came from the mottled, not the uppennost, part of a deep-weathering profile; that the confused associations of the word laterite are such as to cause some workers to recommend its complete abandonment; and that the description of highly siliceous deep-weathering profiles in Australia as lateritic has led to highly unlikely hypotheses of their origin (for discussion of this last topic, see Langford-Smith and Dury, 1965). What may be regarded as a first approximation to the classification of duricrusts originated with Lamplugh (1902, 1907) who introduced the terms calcrete, silcrete, and ferricrete. To these have subsequently been added gypcrete and salcrete. A parallel list replaces -crete by -crust (Table I, cols. i. ii). It seems unnecessary here to identify the points of introduction or the original authorship of the additional terms: for present purposes, they may be accepted as in widespread use, their currency being taken as a guarantee that they are thought useful. Terms in -crete and terms in -crust are not wholly interchangeable. As Wopfner and Twidale (1967) point out, it is appropriate to distinguish between well-cemented and massive gypcrete and loosely-cemented to powdery and impure gypseous crusts. Furthermore, the various -cretes can occur outside the range of -crusts: ortstein for instance can be regarded as a form of ferricrete. A weakness of applying -crete to the material of duricrusts is that the element refers merely to agent of cementation and not necessarily to major bulk content. Thus, while calccrusts are assumed to be composed largely of calcium carbonate (under analysis, 90% or more is common), calcretes need not be. A ready example is provided by the calcite-cemented outwash gravels in the north of the London Basin, where most of the material involved consists of sHica in the form of flint. A general defect in both lists is that they make no provision for highly aluminous duricrusts, presumably because a quite small iron content in the ferricrust range can result in a strong hematitic coloration (cf. however Maignien, 1958, who distinguishes between aluminous and ferruginous latedtes). At the same time, the separation of silcrusts from ferricrusts is useful, as recognising the occurrence of highly siliceous duricrusts, and as superseding the block term lateritic. It is probably too late to debate whether or not calccrusts, gypcrusts, and salcrusts should be classed as duricrusts at all, since the term duricrust appears to be extending itself to embrace them: there is no purpose in trying to turn back a linguistic tide. Equally, the force of usage is probably strong enough to' prevent the substitution of calccrete, the logical accompaniment of calccrust, fDr the original calcr.ete. There can however" be- no" objection to styling crusts in the si1crust- ferricrust range of the first approximation duricrusts proper (cf. Table 1). The difference of origin between duricrusts proper and other types of crust needs constantly to be borne in mind. The latter are associated withsubhumid to arid climates, whether those of the present or those of the past. Many of them have strong affinities with evaporites, even when they do not result directly from the drying of pluvial lakes or from the evaporation of playas or of emergent saline groundwater. However, the name evaporation crusts is not entirely suitable: fOif example, Coque (1962) has identified crusts formed of wind,.laid gypsum. Again, calccrusts can alternatively originate as calcareous soil horizons, including calcareous duripans (7th Approximation, 55-59). The fact that crusts of this subtype form mainly below the level of the ground surface is no necessary obstacle to their classification as crusts, since duricrusts proper originate, at least in very· large part, in precisely the same manner. Gypsic. and for that matter natric and salic soil horizons also (7th Approximation, 45-46. 59-60) seem at most to rise to the status of fragipans (ibid., 56-57) not resisting

78 stripping sufficiently to constitute crusts: in the present context, therefore, they may be disregarded. In respect of calccrusts, it is not prDposed here to review the contrOoversial aspects of the origin of soil carbonate, Oor to discuss the various names which it has been given (e.g., caliche. kunkar. travertine). The tenn calcsilitic crust, mentioned here in anticipation of an outline of the second approximation. is prDvisionally inserted in the classification as connoting calccrust or other surficial limestone such as calcarenite, which has undergone silicification. Investigation is required both into the status of this type and intO' its place in the classification of silicified limestone in general. Duricrusts proper are called, by a number of authors, weathering crusts, but this usage is not wholly defensible. The major extent of duricrusts prDper admittedly seems to' be associated with deep-weathering profiles, of which the duricrusts fOorm part; but scarpfoDt, detrital, and valley-bottom crusts are well documented from numerDUS areas (cf. de Swardt, 1964; Maignien. ap. cit. 1966). where they can either be assDciated with deep weathering or merely result from the cementation of colluvium, terrace deposits, and valley fills. cDnstituting the forms widely called detrital laterites (or lateritites) and ground water laterites. Much obviously depends on the highly variable interplay Df weathering. vegetation. climate, erosion. and time. In this connection. attention may be drawn to a pDtential source of specific confusion. With some writers. primary laterites are those of a residual erosion-platform, secondary laterites those formed below the level of a wasting duricrust cap, and fDrmed from material fallen or leached from it; with others (e.g., Mohr and van Baren. 1954), primary laterite is the material formed next to. weathering bedrock. Confusion of a different kind afflicts the description of duricrusts which do form parts of deep-weathering profiles. Whitehouse (1940) uses pallid and mottled zones for Walther's earlier German names bleichzone, fleckenzone; it has become quite general in Australian aCCDunts of the duricrust to identify the layer above the mottled zone as the indurated zone, so that deep-weathering prDfiles are customarily subdivided into pallid. mottled. and indurated. However, the linguistic structure of this subdivision is Dbviously inconsistent. and can be misleading. In SDme prDfiles, the mottled zone i'S partly or even wholly cemented. and can be both thicker and better cemented than the layer next above; in consequence, it must be classed as part of the duricrust.

The Second Approximation In the construction Df this ratiDnal-descriptive classification, the following principles have been observed :- 1. The basis of classification is bulk content, by weight. 2. Names are indicative of bulk cDntent. 3. Names carry nO' genetic implications. 4. Mineralogy is disregarded.

Application of these principles results in a more detailed classification of duricrusts than has hitherto been avaHable in English. The writer is of course aware Df the work of d'Hoore, (1954), Maignien (op. cit 1958), and others on the classification of laterite, which has been published very largely in French (for a summa,ry review, see Maignien, ap. cit. 1966, 112-113). He is also aware of the considerable advances in genetic interpretation and classification which are involved. However, it can be suggested, without the least hint of d~rogating this wDrk, that certain geomorphic purposes can be well served by a descriptive, as opposed to' a genetic, classification. Particularly is this so in Australia, where duricrusts are widespread and extensive, but where the mechanisms of their

79 Table 1. Classification of Duricrusts. ii iii iv v First Approximation Second Approximation Classification By Essential Chemistry Typical Crystalline Minerals (list not exclusive) (list not exclusive) Cementing Dominant Dominant Agent Content A Content B t t t t silitic crust Si02 quartz rIl rIl Q) Q) ~ ~ siallitic crust Si02, Al20s quartz (aluminous compounds often ~ t ~ t amorphous to cryptocrystalline) rIl rn J, J, fersilitic crust Fe20s, Si02 hematite, quartz 1-; 1-; Q) Q) P. P. 0 0 fersiallitic crust Fe20S, FeOOH, Si02, hematite, goethite, quartz, gibbsite, 1-; 1-; p. p. A120a. nH20, AIOOH 00 + + boehmite 0 rIl rIl !3 Q) !3 .....Q) ~ rIl Q) ferrallitic crust Fe20S, FeOOH, hematite, goethite, gibbsite, boehmite 1-; ;:::l 1-; 1-; eu u u .~ AbOs. nH20" AIOOH ·c ·c ·c I: ;:::l ~ ;:::l Q) "Cl 4-< "Cl 4-< ferritic crust Fe20S, FeOOH, hematitc, goethite

fermangitic crust Fe20S, Mn02 hematite, pyrolusite/psilomelane

tiallitic crust Ti02, Ab03. nH20 rutile/anatase, gibbsite

J, J, J, J, allitic crust AbOa· nH20, AIOOH gibbsite, boehmite

calcitic crust CaCOs calcite ca1crete ca1ccrust { ca1csilitic crust CaCOs, Si02 calcite (silica often chalcedonic, etc.) gypcrete gypcrust gypsitic crust CaS04.2H20 gypsum

salcrete salcrust halitic crust NaCI (usually impure) rock salt formation remain in doubt, not to say in dispute. And particularly also is this so in respect of writings in English, where linguistic precision is less highly esteemed, and less carefully cultivated, than in writings in French. Classification by bulk content raises no problems in the range of crusts other than duricrusts proper. Calccrusts of the calcitic kind commonly run at more than 90% CaCog, and often exceed 95%. TIle proportion of gypsum in gypcrusts is similarly high. Sa1crusts are probably often impure, but surficial occurrences consisting dominantly of rock salt are at least known. Moreover, the allocation of names in this range produces no difficulty (Table 3, col. Hi).

Within the range of duricrusts proper, difficulties at once appear when attempts are made at mineralogical definition, partly because of the variations produced by varying degrees of hydration, partly because of the problems of classifying clay minerals, partly because of the dubiety of mineralogical identification in part of the range, and partly because iron compounds convert to hematite when sufficiently heated, as in analytical practice. These various causes of trouble are sufficiently well known to require no further discussion here (see, among the very numerous available references, Oades, 1963; Maignien, op. cit. 1966), and the mineralogical contents listed in Table 1 are intended merely as the roughest of appro:ximations to actuality.

Particular emphasis may suitably be placed, for purposes of classifiGation, on bulk content of siHcon (as Si02 ), of aluminium (as AI 20 3 ), and or iron (as Fe20a). These fractions commonly supply more than 85%, and indeed more than 95%. of the total weight of a duricrust sample, exclusive of combined H 20. This latter characteristically ranges from about 5% to about 25% of the whole. Locally, enrichment by manganese (as Mn02 ) or by tj,tanium (as Ti02 ) can run up to 20% of the total weight, combined water again excepted. Manganese enrichment is typically associated with low terrace or valley-bottom situations, titanium enrichment with highly leached, residual aluminous duricrusts, the contrast in situation appearing to correspond to a contrast in relative mobility. But the relative proportions of Si02, AI20 g, and Fe20 3 in the majority of duricrusts proper appear to be controlled by an interplay of factors among which relative mobility is only one. Among the most obvious of the others are the constitution of weathering bedrock, the hydrologic, pedologic, biotic, and climatic environments, and the duration of time (cf. Jenny, 1941). By observation, the relative proportions of silicon, aluminium, and iron are highly variable. ranging up to 99% Si02 in some crusts. and up to well above 80% Al20 3 or Fe20a in others. It seems necessary, therefore, to provide descriptive names which allow for the possible sevenfold dominant combinations. Such names are Hsted in Table 1, col. Hi. The names are compounded from the si of silica, al of aluminium, fe of ferrum, and the ti of Htanium. Manganese, for the sake of euphony. is made to supply the element mango The resulting terms of course derive ultimately from the practice of Suess who. in the late 1800s coined the words sal (sial with later writers), and sima, and more nearly from the practice of pedologists. Del Villar (1937, p. 32) for instance appears originally to have applied the names allitic and siallitic to soil series; ferritic, ferrallitic, and fersiallitic now have wide currency among tropical pedologists, especially in Africa, thanks to the work and writings of Aubert, da Costa, d'Hoore, Duchaufour, Maignien, Rougerie, and others. As before, it seems enough to accept the currency of these terms without treating their origin as a problem in historicism (see, however, Aubert, 1954, and references in Maignien, ap. cit. 1966). Their applicability in the classification of duricrusts appears to be self-demonstrating. The term fermangitic, tiallitic, and silitic are thought to be original to the present paper, but it would be in no way surprising if they had already been invented by others: the terms allitic and siallitic had

81 Figure 1. Ternary diagram, showing points for the data in Table 2, and suggesting boundaries among the seven types of duricrust in the total fersiallitic range. already been formed before the writer located them in del Villar (op. cit). All three of fermangitic, tiallitic, and silitic appear to be necessary, especially the last. In respect of it, o"ne could argue that the I should be doubled, by analogy ." with the usage in allitic, producing the form sillitic, but this form appears to be potentially misleading, so that silitic is to be preferred. To draw up a list of c1assificatory names related to proportional content, before criteria of content have been established, may at first seem a reversed exercise in logic. However, the facts are that some names for different kinds of duricrust already exist, and that, as has just been pointed out, names developed in pedology can readily and usefully be adapted to the descriptive identification of duricrusts. The development of language, even of technical language" is little susceptible of control. Nevertheless, it remains to show that the names proposed can be justified by the results of analysis. Analyses of crusts in the siliceous-aluminous-ferruginous range can readily be made to. provide the data for a ternary diagram, when the percentages of Si02 , AlzOg, and FezOg are recalculated so that their sum = 100 (Table 2, Fig. 1). The potential range is, needless to say, any combination among and up to Si,AI,Fe = 100. However. if. as in Figure 1, the 80% limit of a single constituent is used to separate off silitic, allitic, and ferritic crusts, then the margins of the

82 Table 2. Percentage Contents of Si02, Al203 and Fe203 in selected duricrust samples. For sources and locations, see Appendix.

Sample No. 2 3a 3b 4 5 6 7 8 Si02 2.0 3.1 0.9 1.0 7.5 0.5 7.2 3.1 2.4 4.7 After ignition loss Al203 11.0 11.0 49.5 48.0 57.6 58.6 80.4 79.3 Fe203 87.0 84.5 82.6 42.0 41.0 35.0 30.6 7.5 2.4 { Total 93.7 98.6 94.5 92.5 96.5 93.1 96.4 91.0 84.1 Recalculated to Si02 2.0 3.1 1.0 1.1 7.8 0.5 7.5 3.4 2.9 { A1203 5.0 11.2 11.6 53.5 49.7 61.9 60.8 88.4 94.2 ~ = 100 \.. Fe,303 93.0 85.7 87.4 45.4 42.5 37.6 31.7 8.2 2.9 ......

Sample No. 9a 9b 10 11 12 13 14 15 16 Si02 74.5 96.1 79.5 97.8 98.6 98.5 58.7 37.9 33.2 00 w 22.0 2.1 18.4 1.2 0.6 0.7 7.9 9.0 5.0 After ignition loss Al203 Fe203 2.1 1.1 1.1 0.8 0.3 0.6 31.6 40.7 60.2 { Total 98 .6 99.3 99.0 99.8 99.5 99.8 98.2 87.6 98.4 Recalculated to Si02 75.6 96.8 80.4 98.0 99.1 98.7 59.8 43.5 33.7 Ab0 3 22.3 2.1 18.5 1.2 0.6 0.7 8.0 10.0 5.1 :E = 100 { Fe203 2.1 1.1 1.1 0.8 0.3 0.6 32.2 46.5 61.2

Sample No. 17 18 19 20 21 22 23 24

Si02 28.1 23.3 18.9 19.4 34.5 19.8 36.5 28.5 9.1 16.1 27.1 23.6 21.1 32.8 26 .2 44.5 After ignition loss Ab03 Fe203 52.1 60.3 52.0 45.6 42.4 42.0 30.8 26.8 { Total 89.3 99.7 98.0 88.6 98.0 94.6 93.5 99.7 Recalculated to Si02 31.6 23.4 19.2 21.9 35.2 20.9 39.0 28.6 AI203 to.O 16.1 27.7 26.6 21.5 34.7 28.0 44.5 ~ = 100 { Fe203 · 58.4 60.5 53.1 51.5 43.3 44.4 33.0 26.9 remammg field can be separated Ooff for siallitic, fersHitic. and ferrallitic duricrusts, the centre being allocated to fersiallitic crusts. It is in the nature of the geometry Oof the diagram that the subdivisiOon cannot be wholly satisfying. Thus, allitic crusts, as defined, contain up to 20% Si02 + Fe203, and allOow for Si02 = 0%, Fe203 = 20%, and vke versa, whereas the suggested delimitation of siallitic crusts reduces the permitted Fe20 g content to a maximum Oof 10%. Corresponding effects are produced elsewhere. Nevertheless, the sevenfold subdivision of the fersiallitic range can be claimed to cOorrespond far mOore closely to actualities than does the twofOold subdivision of the first approximatiOon. It may be provisionally suggested that the cutoff points for tiaIlitic and fermangitic crusts should also be placed, respectively, at 10% of Ti02 and at 10% of

Mn02 • Some of the data in Figure 1 have been recast, since some workers recOord quartz and other silica separately, while others give quartz as a part of total silica. Again, some of the Ooriginal lists include water of combinati,on, others exclude it: it is excluded frOom Table 2, the values being, where necessary, re-calculated accordingly. All values listed have been corrected tOo the first decimal place. The selected information in Table 2 is meant to shOow that all seven types of duricrust named within the total fersiallitic range can be matched by actual examples. This proves to be so, even during a swift preliminary sorting through the literature. The one near-exception concerns the siallitic type, for the Australian deep-weathering profiles which include highly siliceous duricrusts appear characteristically to be highly siliceous throughout, the sill,ca content in the crust running above 80% and nOot infrequently at abOout 98%. Thus, Joplin (1963, Tables W2a, W4 , pp. 384 - 386, 393) lists some twenty analyses of grey billy (=silcrust) and silcrete, in which the Si02 content is at least 95% in all cases. It has however been thought preferable to rely here on the analysis of samples which have been studied in the specific context of duricrusting, and for which the precise location in the profile is knOown. In some deep-weathering profiles which include highly siliceous crusts, the Al20 g content fails to increase significantly until the pallid zone is reached; its increase there is presumably referable to the layer silicates of kaolin and allied clay minerals. Results from some prOofiles which have been sampled and analyzed in detail fail to suggest that the increase in Al20 s content with depth is in any way orderly; but they do leave open the possibility that at least some siaIlitic crusts result from the cementing of the mottled zone, and the weakness (or weakening) and stripping of the superjacent layer. A more general consideration is that the identification of silitic, fersilitic, and siallitic crusts, hOowever marginal these last may appear in the classificati,on, shows that the relative mobility principle Si> Fe> Al is but one control Oover constitution. The least aluminousfersilitic sample, the very slightly ferruginous siallitic sample, and the most highly ferritic and siIitic samples of thOose plotted in Fig. 1 all come from fossil crusts, and from interfluve sites where relatively great removal of the more mobile constituents might have been looked fOor. The identification in the literature of the deduced types of duricrust proper which are listed in Table 1 is a the same time a confirmation of the efficacy (in this context) of the deductive process, and a suggestion of possible directions for further investigation in the future.

84 ACKNOWLEDGMENTS I am indebted to T. Langford-Smith. Associate Professor of Geomorphology. The University of Sydney. for communicating the results of the analysis of a number of Australian samples of duricrust. These results have not previously been recorded in print. The collection of the samples. and their analysis. were undertaken as part of our joint work which is financed by the Australian Research Grants Committee. Project Reference No. B65/15475.

APPENDIX

Location of Samples: Proximate Sources of Analytical Data. Among the sources, Barshad is Chap. 1 in Bear (ed.) (1964); the other publication references are to Maignien (1966) and Mohr and van Baren (1954). Sample Location and Proximate Source No. 1. Coolgardie, W. Australia (Barshad, p. 126). 2. Kaloum Peninsula, Guinea (Maignien, p. 38). 3a,3b. Eagle Mountain, Guyana (Mohr and van Beran, p. 134). 4. Mt Bougourou, Guinea (idem., p. 151). 5. Sarata, Bombay, India (Barshad, p. 126). 6. Iles de Loos, Guinea (Maignien, p. 38). 7. Ile de Roume, Guinea (Mohr and van Baren, p. 151). 8. Bihar, India (Barshad, p. 126). 9a. Near Quilpie, Qld, Australia: surface (unpublished). 9b. Same profile as 9a, c. 5ft below surface, but still above mottled zone (unpublished) . 10. Near Thargomindah, Qld, Australia (unpublished). 11. Near Wilcannia, NSW, Australia (unpublished). 12. Near Thargomindah, Qld, Australia (unpublished). 13. Near Roma, Qld, Australia (unpublished). 14. Near Gulgandra, NSW, Australia (unpublished). 15. Siam, from structure (Mohr and van Baren, p. 378). 16. Near Mitchell, Qld, Australia (unpublished). 17. Siam, field sample (Mohr and van Baren, p. 378) . 18. Near Miles, Qld, Australia (unpublished). 19. Mamou, Guinea (Barshad, p. 126). 20. Cheruvannar, India: Buchanan's original 1807 site (idem., p. 126) . 21. Dahomey (idem., p. 126). 22. Tumatumari, Guyana (Mohr and van Baren, p. 140) . 23. Yarikita Hill, Guyana (idem., p. 143). 24. Near Charleville, Qld, Australia (unpublished) ,

REFERENCES

Aubert, G., 1954: Les Sols Lateritiques. Trans. 5th Internat. Congr. Soil Sci., Leopoldville, Vol. I, pp. 103-119. Bear, F. E. (ed.) , 1964: Chemistry of the Soil. New York: Reinhold Publishing Corpn. Coque, R., 1962: La Tunisie Presaharienne. Paris: Armand Colin. Jenny, H., 1941 :Factors of Soil Formation - A System of Quantitative Pedology. New York: McGraw-Hil1. Joplin, G. A., 1963: Chemical Analyses of Australia Rocks, Part I. Cmnwlth Australia Dept. Natl. Develpt., Bur. Min. Resources Bulletin, No. 65 d'Hoore, J., 1954: L'accumulation des Sesquioxides Libres dans les Sols Tropicaux. Brussels: INEAC, Ser. Sci. No. 62. Lamplugh, G. W., 1902: "Calcrete." Geol. Mag. Dec. 4, Vol. 9, p. 575. -----" 1907: The Geology of the Zambesi Basin and Batoka Gorge (Rhodesia). Quart. J. Geol. Soc. London, 63, pp. 162-216. Langford-Smith, T., and Dury, G. H., 1965: Distribution, Character and Attitude of the Duricrust in the Northwest of New South Wales and the Adjacent Areas of Queensland. Amer. J. Sci., 263, pp. 170-190. Maignien, R., 1958: Le Cuirassement des Sols en Guinee. Strasbourg: Mem. Carte. Geo!. Alsace-Lorraine, No. 16. -----, 1966: Review of Research on Laterites. Paris: UNESCO, Natural Resources Research Ser., IV.

85 Mohr, E. C. l., and van Baren, F. A., 1954: Tropical Soils. London and New York: Inter- science Publishers. Oades, l. M., 1963: The Nature and Distribution of Iron Compounds in Soils. Soils and Fertilisers, 26, pp. 69-80. Prescott, l. A., and Pendleton, R L., 1952: Laterite and Lateritic Soils. Farnham Royal: Cmnwlth Bur. Soil Sci. Tech. Comm. No. 47. de Swardt, A. M. J., 1964: Lateritization and Landscape Development in Parts of Equatorial Africa. Zeit. fur Geomorph., 8, pp. 313-333. Thornthwaite, C. W., 1931: The Climates of North America according to a New Classifica- tion. Geog. Review, 21, pp. 633-655. -----, 1948: An Approach toward a Rational Classification of Climate. ibid., 38, pp. 55-94. United States Department of Agriculture, 1960: Soil Classification, A Comprehensive System, 7th Approximation. Washington, D.e.: U.S. Govt. Printing Office. del Villar, E. H., 1937: Los Suelos de la Peninsula Luso-Iberica (ed. G. W. Robinson). Madrid; Author, and London: Thomas Murby & Co. Whitehouse, F. W., 1940: Studies in the Late Geological History of Queensland. Dept. of Geol., U. of Qld., Papers, Vo!. 2 (New Ser.) , No. 1, pp. 1-74. Woolnough, W. G., 1927: The Duricrust of Australia. I. Proc. Roy. Soc. N.S.W., 61, pp. 24.53. Wopfner, H., and Twidale, C. R, 1967: Geomorphological History of the Lake Eyre Basin. Pp. 95-117 of Landform Studies from Australia and New Guinea (eds., J. N. Jennings and J. A. Mabbutt), Canberra: A.N.V. Press.